Genetic Variation in the TGF-β Signaling Pathway and Colon and Rectal Cancer Risk Free

The TGF-β signaling pathway is an essential regulator of cellular proliferation, differentiation, apoptosis, and extracellular matrix remodeling in the cell (1). In addition, this signaling pathway is involved in angiogenesis and inflammation. It mediates intracellular actions of proinflammatory cytokines, including activation of nuclear factor-kappa B (NFκB; refs. 2, 3) and deficiency of TGF-β has been shown to lead to extensive inflammation (2). TGF-β ligands initiate their cellular effects by binding to cell surface receptors (1); type 1 receptors mediate their cellular effects through interaction with Smad proteins. Thus, Smads are key intracellular mediators of the transcriptional responses to TGF-β (4).

Smad4 (DPC4) is inactivated in some colorectal cancers (CRC), and germline mutations of Smad4 have been linked to familial juvenile polyposis families (5). Smad2 has been identified as a TGF-β–responsive Smad that is a transcription factor involved in the regulation of cell growth and apoptosis. Smad7 is also involved in inflammation-related pathways and has been shown to modulate TGF-β and Wnt signaling (6). Genetic variation in the Smad7 gene on 8q21 has been identified through numerous genome-wide association studies (GWAS) as being associated with CRC (7). Like Smad7, Smad2 and Smad4 are located on 8q21. We previously reported on the replication of tagSNPs in the Smad7 gene identified from GWAS in our population-based case–control study of colon cancer (8). We observed that rs12953717 was associated with a statistically significant increased risk of colon cancer [odds ratio (OR) = 1.38, 95% CI = 1.13–1.68; P linear trend < 0.01) for the TT genotype compared with the CC genotype whereas the CC genotype of the rs4939827 tagSNP was inversely associated with colon cancer (OR = 0.77, 95% CI = 0.64–0.93) relative to the TT genotype. In our study, associations seemed to be modified by use of aspirin (8).

There is growing support for the role of the TGF-β signaling pathway in the etiology of colon and rectal cancer. In this study, we evaluate genetic variation in TGFβ1, TGFβR1, Smad1, Smad2, Smad3, and Smad4. We evaluate how these genes interact with other potentially important genes in the pathway, including Smad7, NFκB1, and IKBkB, involved in inflammation-related mechanisms. Environmental factors that may operate in this pathway include estrogen, aspirin/non-steroidal anti-inflammatory drugs (NSAID), and cigarette smoking that may lead to oxidative stress and increase the likelihood of inflammation (9). We evaluate the potential interactions between these factors and genetic variation in the TGF-β signaling pathway. In addition, we seek to confirm previous reports that genetic alterations in the TGF-β signaling pathway influences tumor markers such as microsatellite instability and epigenetic changes. We evaluate the hypothesis that the TGF-β signaling influences prognosis after diagnosis with cancer by comparing survival rates based on genetic variation in this pathway.

Two study populations are included in these analyses. The first study, a population-based case–control study of colon cancer, included cases (n = 1,593) and controls (n = 1,994) identified between October 1, 1991, and September 30, 1994 (10), living in the Twin Cities Metropolitan Area, Kaiser Permanente Medical Care Program of Northern California (KPMCP), and a 7-county area of Utah. The second study, with identical data collection methods, included cases with cancer of the rectosigmoid junction or rectum (n = 790) and controls (n = 999) who were identified between May 1997 and May 2001 in Utah and KPMCP (11). Eligible cases were between 30 and 79 years old at time of diagnosis, English speaking, mentally competent to complete the interview, had no previous history of CRC, and no known (as indicated on the pathology report) familial adenomatous polyposis, ulcerative colitis, or Cohn's disease.

Controls were matched to cases by gender and by 5-year age groups. At KPMCP, controls were randomly selected from membership lists; in Utah, controls 65 years and older were randomly selected from the Health Care Financing Administration lists and controls younger than 65 years were randomly selected from driver's license lists. In Minnesota, controls were selected from driver's license and state-identification lists. Study details have been previously reported (12, 13).

Data were collected by trained and certified interviewers using laptop computers. All interviews were audio-taped as previously described and reviewed for quality control purposes (14). The referent period for the study was 2 years prior to diagnosis for cases or selection for controls. Detailed information was collected on diet, physical activity, medical history, reproductive history, family history of cancer in first-degree relatives, regular use of aspirin and NSAIDs, and body size.

Tumor registry data were obtained to determine disease stage at diagnosis and months of survival after diagnosis. Disease stage was categorized by Surveillance, Epidemiology, and End Results (SEER) staging of local, regional, and distant disease as well as by the American Joint Committee on Cancer (AJCC) staging criteria. Local tumor registries provided information on patient follow-up including vital status, cause of death, and contributing cause of death. Survival months were calculated on the basis of month and year of diagnosis, and month and year of death, or date of last contact for those individuals who were still alive.

We have previously evaluated tumors for CpG island methylator phenotype (CIMP), microsatellite instability (MSI), TP53 mutations, and KRAS2 mutations (15–18) and were therefore able to evaluate genes in relation to tumors with specific characteristics or markers. Details for methods used to evaluate these epigenetic and genetic changes have been described in previous publications (15–18).

tagSNPs were selected for genes TGFβR1, Smad1, Smad2, Smad3, and Smad4, using the following parameters: an r² < 0.8 defined LD blocks using a Caucasian LD map, minor allele frequency or MAF > 0.1, range = −1,500 bp from the initiation codon to +1,500 bp from the termination codon, and 1 SNP/LD bin. All markers were genotyped using a multiplexed bead array assay format based on GoldenGate chemistry (Illumina). A genotyping call rate of 99.85% was attained. Blinded internal replicates represented 4.4% of the sample set. The duplicate concordance rate was 100.00%.

For TGFβ1, candidate markers rs1800469 and rs4803455 were chosen on the basis of prevalent MAF and previous findings described in the literature (19); rs1800469 and rs4803455 were genotyped independently using a TaqMan assay from Applied Biosystems. Each 5-μL PCR reaction contained 20 ng of genomic DNA, primers, probes, and TaqMan Universal PCR Master Mix (containing AmpErase UNG, AmpliTaq Gold enzyme, dNTPs, and reaction buffer). PCR was carried out under the following conditions: 50°C for 2 minutes to activate UNG, 95°C for 10 minutes, followed by 40 cycles of 92°C for 15 seconds, and 60°C for 1 minute using 384 well duel block ABI 9700. Fluorescent endpoints of the TaqMan reactions were measured using a 7900HT sequence detection instrument.

All statistical analyses were carried out using SAS version 9.2 (SAS Institute). We assessed ORs and 95% CIs in multiple logistic regression models for colon and rectal cancer separately. All SNPs were evaluated first by comparing the heterozygote and homozygote variants to the homozygote wild-type and subsequently assessing the likelihood of the dominant and recessive models of inheritance; the best fitting model is presented (20). P values from the unadjusted Max test were used to adjust for multiple comparisons of tagSNPs using the methods by Conneely and Boehnke (20, 21). Minimal adjustments were made for age, sex, race, and study center. Additional adjustments for BMI (kg/m²), physical activity, use of aspirin or NSAIDs within 2 years of the referent period, and cigarette smoking status (ever or never regularly smoked) did not alter associations.

Stepwise regression models were used to identifying tagSNPs that contributed uniquely and most significantly to the overall fit of the model for colon and rectal as well as to identify potential confounding of tagSNPs within genes. Inclusion in the stepwise regression model was based on a score chi-square significance level of 0.05 whereas exclusion was determined on the basis of a Wald chi-square 0.05 significance level. Subsequent analysis for interaction was based both on tagSNPs remaining in the final stepwise model and those identified as being important independently.

We evaluate interaction between TGFβ1 and its receptor and Smad1, Smad2, Smad3, Smad4, Smad7, IKBκB, and NFκB1. Possible interactions between SNPs and sex, age (30–64 or 65–79), recent aspirin or NSAID use, estrogen status, BMI (<25, 25–30, >30), and cigarette smoking were evaluated given the hypothesized mechanisms proposed for these genes. Associations between colon cancer and Smad7, IKBkB, and NFκB1 have been previously reported (8; K. Curtin, R.K. Wolff, J.S. Herrick, R. Abo, M.L. Slattery, unpublished data). P values for interaction were determined by comparing a full model including an ordinal multiplicative interaction term to a reduced model without an interaction term using a likelihood ratio test; a categorical model was used for TGFβ1 rs4803455 and smoking and for Smad2 rs1792689 and TGFβR1 rs1571590. Haplotypes based on the SNPs being identified as significant for each gene were examined with both environmental and gene interactions but did not yield any more meaningful results than looking at the individual SNPs and therefore are excluded.

Tumors were defined by specific alterations detected; any TP53 mutation, any KRAS2 mutation, MSI+, or CIMP+ defined as at least 2 of 5 markers methylated. As the proportion of MSI+ tumors in the rectal cases was <3% (22), there was insufficient power to examine these tumor markers with genotype data. Population-based controls were used to assess associations for the population overall when examining multiple outcomes defined by tumor status. In addition to identifying variants that contributed to a given phenotype independently, a stepwise regression of all SNPs per gene was implemented in SAS, using the logistic procedure for each individual tumor type.

Time of survival was determined on the basis of date of diagnosis and date of last contact or death, truncated at 5 years, the time period which is most meaningful for assessment of impact with CRC. Associations between SNPs and risk of dying of CRC within 5 years from diagnosis were evaluated using Cox proportional hazards models to provide multivariate hazard rate ratios (HRR) and 95% CIs adjusted for age at diagnosis, study center, race, sex, AJCC stage, and tumor markers. HRRs were assessed for SNPs independently and using stepwise regression via the phreg procedure adjusting for other SNPs.

Table 1 describes the genes and corresponding SNPs associated independently, through interaction, or with tumor markers. All SNPs were in Hardy–Weinberg equilibrium (HWE). SNPs that were independently associated with colon or rectal cancer overall are shown in Figure 1. As shown in the figure, the following associations were observed for colon cancer: OR = 1.25 (95% CI = 1.03–1.51) TT versus AA for Smad2 rs1787199; OR = 1.33 (95% CI = 1.06–1.67) CC versus TT for Smad2 rs4940086; OR = 0.68 (95% CI = 0.55–0.85) for AG/GG versus AA for Smad3 rs12901071; OR = 0.69 (95% CI = 0.57–0.84) CC versus AA for Smad3 rs1498506; OR = 0.76 (95% CI = 0.59–0.98) for AA versus GG/GA for Smad3 rs7163381, adjusted for rs1498506; OR = 0.68 (95% CI = 0.47–0.97) CC versus GG/GC for Smad3 rs2414937; OR = 0.65 (95% CI = 0.51–0.84) for AA versus GG for TGFβ1 rs1800469; OR = 1.43 (95% CI = 1.18–1.73) for AA versus CC for TGFβ1 rs4803455; OR = 0.85 (95% CI = 0.74–0.99) for TA/AA versus TT for TGFBR1 rs6478974. After adjustment for multiple comparisons, Smad3 rs1498506 and rs12901071 remained statistically significant (adjusted P values of 0.0.009 and 0.015, respectively). Because TGFβ1 rs1800469 and rs4803455 were candidate SNPs, we did not adjust them for multiple comparisons.

The following associations were statistically significant for rectal cancer (Figure 1): OR = 0.78 (95% CI = 0.62–0.98) for CT/TT versus CC for Smad2 rs1792689; and OR = 1.81 (95% CI = 1.12–2.91) for CC versus TT/TC for Smad3 rs17293443. Although Smad3 rs11071933 and rs1866317 were not statistically significant independently, after adjusting for rs17293443 and one another, risk estimates were 0.75 (95% CI = 0.61–0.93 and 1.28 (95% CI = 1.03–1.59) for the CG/GG versus CC genotypes, respectively.

For colon cancer, we observed a statistically significant interaction between Smad3 rs3825977 and TGFβ1 rs1800469 and between Smad2 rs4940086, Smad3 rs17293443, and Smad7 rs4939827 with TGFβ1 rs4803455 (Table 2). Statistically significant interactions also were observed between both TGFBR1 rs6478974 and rs1571590 with IKBkB rs37473811 and with NFκB1 rs4648110 (Table 2). Statistically significant gene–gene interactions also were identified for rectal cancer (Table 3). TGFβ1 rs1800469 interacted with Smad3 rs211860 and rs4147358 (Table 3); TGFβ1 rs4803455 and TGFβR1 rs105733710 interacted with NFκB1 rs4648110 and rs13117745; TGFβR1 rs1571590 interacted significantly with Smad2 rs1792689.

Several variants within the TGF-β signaling pathway interacted with lifestyle factors hypothesized as influencing this pathway. Statistically significant interactions with cigarette smoking and colon cancer were observed for TGFβ1 rs4803455, TGFβR1 10733710, and rs1571590 (Table 4). As previously noted, the AA genotype of TGFβ1 rs4803455 increased risk of colon cancer overall, but the increase in risk was especially dramatic among recent smokers (OR = 2.09, 95% CI = 1.47–2.96). The GG genotype of TGFβR1 rs1571590 was associated with increased colon cancer risk among nonsmokers/former smokers while there was a trend toward reduced risk among recent cigarette smokers for the same genotype. The A allele of TGFβ1 rs1800469 was observed as increasing rectal cancer risk among recent smokers.

The TGFβR1 rs6478974 A allele was associated with reduced risk of colon cancer among those who recently used aspirin/NSAIDs and had no effect among non–aspirin/NSAID users (Table 4). Smad3 rs3743343 interacted significantly with aspirin/NSAIDs for both colon and rectal cancers, although the direction of the association was different for these cancer sites. Statistically significant interactions were observed for Smad3 rs7173811 and aspirin/NSAIDs for colon cancer and both Smad3 rs7163381 and rs11071933 and rectal cancer. Among these SNPs, those who had the variant allele were at increased risk if they did not use aspirin/NSAIDs regularly but were at significantly reduced risk if they used aspirin/NSAIDs regularly.

Among women recently exposed to estrogen, the A allele of TGFβ1 rs1800469 was associated with a reduced risk of colon cancer and the C allele of rs4803455 was associated with a decreased risk of rectal cancer (Table 4). Likewise, both variants of Smad4, rs10502913 and rs8096092, were associated with increased risk of rectal cancer among men while reducing risk among women.

Unique sets of Smad2, Smad3, TGFβ1, and TGFβR1 SNPs were associated with tumor phenotypes for colon and rectal cancer (Table 5). Among colon cancer cases, the risk of a CIMP+ tumor was associated with both Smad2 and Smad3. TGFβ1 rs1800469 was associated with a decreased risk for all colon tumor phenotypes except CIMP+, although not associated with rectal molecular phenotype. TP53-mutated colon tumors were associated with Smad2 rs4940086 and Smad3 rs7176870. MSI+ colon tumors were associated with Smad2 rs1792689 and rs1787199 and Smad3 rs12901071 and rs731874. For rectal cancer, Smad3 rs893473 was associated with an increased likelihood of a CIMP+ tumor (OR = 3.6, 95% CI = 1.62–798) for the TT genotype relative to CC/CT; rs991157 AA versus GG/GA was associated with a statistically significant increased risk of a KRAS2-mutated tumor (OR = 1.63, 95% CI = 1.03–2.79). The TGFβR1 rs10733710 GA/AA genotype was associated with increased risk for both CIMP+ tumors and TP53-mutated tumors.

Variations in TGFβ1, Smad1, Smad2, and Smad4 were not associated with survival after diagnosis (data not shown in table). Four SNPs were associated with colon cancer survival: TGFβR1 rs10733710 GA/AA versus GG (HRR = 0.73, 95% CI = 0.57–0.95); and 3 Smad3 SNPs, rs11639295 TT versus CC/CT (HRR = 0.46, 95% CI = 0.27–0.80); rs12708492 CT/TT versus CC (HRR = 1.78, 95% CI = 1.27–2.50), and rs2414937 CC versus GG (HRR = 2.54, 95% CI = 1.29–3.95). For rectal cancer, 4 SNPs also were associated with survival, although the associated SNPs were different from those that were associated with colon cancer. For rectal cancer, the associations were as follows: TGFβR1 rs6478974 AA versus TT genotype (HRR = 1.73, 95% CI = 1.08–2.78) and rs1571590 AG/GG versus AA genotype (HRR = 0.64, 95% CI = 0.43–0.95); Smad3 rs12904944 GA/AA versus GG (HRR = 1.45, 95% CI = 1.03–2.04) and rs3825977 CT/TT versus CC genotype (HRR = 1.55, 95% CI = 1.10–2.18).

The TGF-β signaling pathway is thought to play a critical role in the carcinogenic process because of its involvement in the regulation of cell growth, differentiation, proliferation, and apoptosis (23). TGF-β exerts its physiologic effect by activating its receptors. Once the TGF-β receptor complex is activated, intracellular signaling is initiated. The TGF-β receptor complex activates the Smad signaling pathway by directly phosphorlyating Smad2 and Smad3 that work in conjunction with Smad4 (24). In this study, genetic variation in TGFβ1 was associated with an increased risk of colon cancer, but not rectal cancer. Our evaluation of genetic variation in TGF-β signaling pathway showed several variants associated with colon and rectal cancer, acting independently as well as modifying the effect of other genetic and lifestyle factors.

A major function of TGF-β is mediating intracellular actions of proinflammatory cytokines, including activation of NFκB (2, 3). Deficiency of TGF-β has been shown to lead to extensive inflammation (2). Inflammation status of the gut seems to play a critical role in the etiology of both colon and rectal cancers (25). Our data support the role of TGF-β in an inflammation-related pathway, given the interaction between genetic variants of NFκB1 and TGFβ1 and TGFβR1 for both colon and rectal cancers. NFκB is an important nuclear transcription factor that regulates a large number of cytokines and is critical for the regulation of inflammation; increased transcription of NFκB can increase inflammation and angiogenesis as well as cell survival and growth (26). IkBκB is a key regulator of NFκB transcriptional activity (27); its proteins are inhibitors of NFκB (26). In addition to the interaction between other genes involved in the regulation of inflammation and variants in the TGF-β signaling pathway, we observed significant interaction with recent use of aspirin/NSAIDs and TGFβR1 rs6478974 and risk of colon cancer, further supporting an inflammation-related mechanism.

It has been hypothesized that cigarette smoking can influence inflammation via enhanced oxidative stress. Furthermore, cigarette smoke has been shown to regulate the effect of various cytokines, including TGF-β (27–30). We observed statistically significant interaction between TGFβ1 and TGFβR1 variants and cigarette smoke and colon cancer, thus supporting this link in a population-based study. We also observed statistically significant interaction between estrogen and TGFβ1 rs4803455. Estrogen has many physiologic properties and has been shown to influence both inflammation and insulin (31, 32).

One of the major mechanisms of TGF-β signaling is through a Smad-dependent pathway (6); Smad7 promotes the anti-inflammatory action of the TGF-β signaling pathway (6). Thus, we evaluated how genetic variants between TGFβ1 and TGFβR1 were associated with Smad2, Smad3, Smad4, and Smad7. We have previously reported on independent associations between Smad7 and colon cancer (8). In this article, we provide information on Smad2, Smad3, and Smad4, which have been hypothesized as important components of the TGF-β signaling pathway (34), as well as evaluate how Smad7 interacts with other genes in the pathway. Both Smad2 and Smad3 showed independent associations with colon cancer; however, several variants also showed consistent associations with CIMP+ tumors. Smad has been associated with epigenetic silencing in other cancers (35). Smad2 and Smad7 interacted significantly with TGFβ1 and TGFβR1, further supporting the importance of multiple elements of the TGF-β signaling pathway in the etiology of colon and rectal cancer.

Both TGFβR1 and Smad3 were associated with survival after diagnosis with colon and rectal cancer. We evaluated genetic variations in our candidate pathway because of its documented role in cell differentiation, metastasis, and survival (36–38). These associations were detected independent of stage at time of diagnosis and tumor characteristics. Although many SNPs were associated with survival, the ones of most importance often varied after diagnosis with colon versus rectal cancer. It is not readily clear why these differences were observed; however, many differences have been detected previously for colon and rectal cancer, suggesting different elements to their etiology and possible prognosis.

There are many strengths and limitations to this study. Others have evaluated polymorphisms in TGFβ1 with CRC and have found some associations with some polymorphisms (39, 40). In our study, we were able to thoroughly evaluate this candidate pathway, using both tagSNP and haplotype analysis, looking at colon and rectal cancers separately and evaluating associations that may be unique to certain tumor molecular phenotypes. The data are extensive and allow us to evaluate interactions with hypothesized genes as well as with hypothesized lifestyle factors. This approach has enabled us to acquire a more comprehensive understanding of the TGF-β signaling pathway and colon and rectal cancer. Although the candidate pathway and specific genes were hypothesize a priori as being associated with colon and rectal cancer, the process of a thorough evaluation lead to many comparisons. Replication of these findings in other studies is therefore needed.

Our data suggest that the TGF-β signaling pathway in conjunction with Smad is an important component of colon and rectal cancer risk and survival after diagnosis. Environmental factors, such as smoking cigarettes and using aspirin/NSAIDs, modulate this risk. Also of importance is the finding that some of these genes preferentially influenced the development of CIMP+ tumors, providing additional information on the carcinogenic process. Support for these findings from other similar studies is necessary to verify these associations.

The contents of this manuscript are solely the responsibility of the authors and do not necessarily represent the official view of the National Cancer Institute.

We thank Drs. Bette J. Caan, Kristin Anderson, and John D. Potter, Sandra Edwards, Roger Edwards, Leslie Palmer, Donna Schaffer, and Judy Morse for data management and collection.

This study was funded by NCI grants CA48998 and CA61757. This research also was supported by the Utah Cancer Registry, which is funded by contract #N01-PC-67000 from the NCI, with additional support from the State of Utah Department of Health, the Northern California Cancer Registry, and the Sacramento Tumor Registry.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.